AI Coding Assistant is a developer-friendly tool that uses cutting-edge LLMs — Gemini and Groq (DeepSeek-LLaMA) — to help users debug, generate, and optimize Python code. It thinks step-by-step and outputs clean, modular, and well-commented Python code while explaining the logic behind it.
- 🧠 Expert-level system prompt with thought-out reasoning
- 💬 Natural language interface using Streamlit chat UI
- 🤖 Supports both Gemini 2.0 Flash and DeepSeek LLaMA 70B
- 📝 Step-by-step
<think>explanations and final code answers - ♻️ Persistent chat history using
st.session_state ⚠️ Clear API key handling using.env- 💡 Beginner-friendly output with in-code comments
| Layer | Technology |
|---|---|
| UI | Streamlit |
| LLM (Option 1) | Google Gemini API |
| LLM (Option 2) | Groq DeepSeek-LLaMA |
| Env Manager | python-dotenv |
| Language | Python 3 |
git clone https://github.com/Md-Emon-Hasan/AI-Coding-Assistant.git
cd AI-Coding-Assistant
pip install -r requirements.txtCreate a .env file in your root folder:
GOOGLE_API_KEY=your_gemini_api_key_here
GROQ_API_KEY=your_groq_api_key_herestreamlit run app.py- User types a Python-related question or error
- The assistant thinks step-by-step within
<think>tags - Outputs a clean, optimized code block
- Both Gemini and Groq models are integrated (one active per app)
AI-Coding-Assistant/
├── .env
├── app_gemini.py # Gemini-based assistant
├── app_groq.py # Groq (DeepSeek) based assistant
├── requirements.txt
└── README.md
streamlit
python-dotenv
google-generativeai
groq| Gemini | Groq |
|---|---|
Made with ❤️ by Md Emon Hasan
- 📧 Email: iconicemon01@gmail.com
- 💬 WhatsApp: Click to Chat
- 💻 GitHub: Md-Emon-Hasan
- 🔗 LinkedIn: Md Emon Hasan
- 📘 Facebook: Md Emon Hasan
This project is open-source under the MIT License.
If this project helped you, please consider giving it a ⭐ on GitHub. It helps a lot!